The overarching goal of this project has been, and continues to be, to develop treatment paradigms for individuals with stroke-induced morphosyntactic language deficits, i.e., agrammatic aphasia. This work makes use of mutually supportive representational (linguistic) and processing accounts of language, as well as psycholinguistic and neurolinguistic research findings, as the basis for understanding language breakdown, selecting training targets, and predicting recovery patterns. Patterns of language (re)learning and generalization provide blueprints for clinical protocols and, in turn, address the utility of this translational approach for studying language disorders. Training involves the use of metalinguistic tasks, which exploit linguistic properties and constructs involved in building grammatical sentences. This approach, termed Treatment of Underlying Forms (TUF), has proved to be successful in our past work, showing that linguistically related structures recover in parallel and that maximal recovery results from training more complex rather than simple structures. Four sets of experiments are proposed for the next grant cycle: Set 1 extends our sentence work from Wh- to NP-movement structures. In Set 2, we continue to examine the relation between and among grammatical morphemes, motivating complexity hierarchies by morphological, rather than syntactic, theories and data. We also experimentally examine the relation between syntax and morphology in Set 3 experiments by training syntactic structures and testing generalization to related grammatical morphemes, and vice versa. Finally, Set 4 experiments focus on verb argument structure, testing recovery of canonical sentence forms by controlling the number and type of arguments selected by the verb. The processing mechanisms that support recovery are studied as part of each set of experiments by tracking eye movements during language comprehension and production tasks and the neural correlates of recovery are examined using functional magnetic resonance imaging (fMRI). Both methods used in the previous cycle revealed important processing patterns as well as, in the case of fMRI, pre- to post-training effects on activation patterns. We, therefore, propose to continue these efforts in this continuation proposal. PUBLIC HEALTH RELEVANCE The overarching goal of this project is to develop treatment paradigms for stroke-induced agrammatic aphasia, a language deficit affecting sentence production (and comprehension) ability. Linguistic theory as well as psycholinguistic and neurolinguistic research findings are used as the basis for understanding language deficit patterns, selecting treatment targets, and predicting recovery patterns. The processing mechanisms that support recovery also are studied by tracking eye movements during language comprehension and production tasks and the neural correlates of recovery are examined using functional magnetic resonance imaging (fMRI).